The relationship between structure and function of molecules is a fundamental issue in the study of biological systems. Structure-function relationships are important in understanding, for example, the function of enzymes, cellular communication, and cellular control and feedback mechanisms. Certain macromolecules are known to interact and bind to other molecules having a specific three-dimensional spatial and electronic distribution. Any macromolecule having such specificity can be considered a receptor, whether the macromolecule is an enzyme, a protein, a glycoprotein, an antibody, an oligonucleotide sequence of DNA, RNA or the like. The various molecules that receptors bind are known as ligands.
Pharmaceutical drug discovery is one type of research that relies on the study of structure-function relationships. Most contemporary drug discovery involves discovering novel ligands with desirable patterns of specificity for biologically important receptors. Thus, the time necessary to bring new drugs to market could be greatly reduced by the discovery of novel methods which allow rapid screening of large numbers of potential ligands.
Since the introduction of solid phase synthesis methods for peptides and polynucleotides new methods employing solid phase strategies have been developed that are capable of generating thousands, and in some cases even millions, of individual peptide or nucleic acid polymers using automated or manual techniques. These synthesis strategies, which generate families or libraries of compounds, are generally referred to as "combinatorial chemistry" or "combinatorial synthesis" strategies.
Combinatorial chemistry strategies can be a powerful tool for rapidly elucidating novel ligands to receptors of interest. These methods show particular promise for identifying new therapeutics. See generally, Gorgon et al., "Applications of Combinatorial Technologies to Drug Discovery: II. Combinatorial Organic Synthesis, Library Screening Strategies, and Future Directions," J. Med. Chem 37:1385-401 (1994) and Gallop et al., "Applications of Combinatorial Technologies to Drug Discovery: I. Background and Peptide Combinatorial Libraries," J. Med. Chem 37:1233-51 (1994). For example, combinatorial libraries have been used to identify nucleic acid aptamers, Latham et al., "The Application of a Modified Nucleotide in Aptamer Selection: Novel Thrombin Aptamers Containing 5-(1-Pentynyl)-2'-Deoxy Uridine," Nucl. Acids Res. 22:2817-2822 (1994); to identify RNA ligands to reverse transcriptase, Chen & Gold, "Selection of High-Affinity RNA Ligands to Reverse Transcriptase: Inhibition of cDNA Synthesis and RNase H Activity," Biochemistry 33:8746-56 (1994); and to identify catalytic antibodies specific to a particular reaction transition state, Posner et al., "Catalytic Antibodies: Perusing Combinatorial Libraries," Trends. Biochem. Sci. 19:145-50 (1994).
The diversity of libraries generated using combinatorial strategies is impressive. For example, these methods have been used to generate a library containing four trillion decapeptides, Pinilla et al., "Investigation of Antigen-Antibody Interactions Using a Soluble, Non-Support-Bound Synthetic Decapeptide Library Composed of Four Trillion (4.times.10.sup.12) Sequences," Biochem. J. 301:847-53 (1994); 1,4-benzodiazepines libraries, Bunin et al., "The Combinatorial Synthesis and Chemical and Biological Evaluation of a 1,4-Benzodiazepine Library," Proc. Natl. Acad. Sci. 91:4708-12 (1994) and U.S. Pat. No. 5,288,514, entitled "Solid Phase and Combinatorial Synthesis of Benzodiazepine Compounds on a Solid Support," issued Feb. 22, 1994; libraries containing multiple small ligands tied together in the same molecules, Wallace et al., "A Multimeric Synthetic Peptide Combinatorial Library," Pept. Res. 7:27-31 (1994); libraries of small organics, Chen et al., "`Analogous` Organic Synthesis of Compound Libraries: Validation of Combinatorial Chemistry in Small-Molecule Synthesis," J. Am. Chem. Soc. 116:2661-2662 (1994); libraries of peptidosteroidal receptors, Boyce & Nestler, "Peptidosteroidal Receptors for Opioid Peptides: Sequence-Selective Binding Using a Synthetic Receptor Library," J. Am. Chem. Soc. 116:7955-7956 (1994); and peptide libraries containing non-natural amino acids, Kerr et al., "Encoded Combinatorial Peptide Libraries Containing Non-Natural Amino Acids," J. Am. Chem. Soc. 115:2529-31 (1993).
To date, three general strategies for generating combinatorial libraries have emerged: "spatially-addressable," "split-bead" and recombinant strategies. These methods differ in one or more of the following aspects: reaction vessel design, polymer type and composition, control of physical constants such as time, temperature and atmosphere, isolation of products, solid-phase or solution-phase methods of assay, simple or complex mixtures, and method for elucidating the structure of the individual library members.
Of these general strategies, several sub-strategies have been developed. One spatially-addressable strategy that has emerged involves the generation of peptide libraries on immobilized pins that fit the dimensions of standard microtitre plates. See PCT Publication Nos. 91/17271 and 91/19818, each of which is incorporated herein by reference. This method has been used to identify peptides which mimic discontinuous epitopes, Geysen et al., BioMed. Chem. Lett. 3:391-404 (1993), and to generate benzodiazepine libraries, U.S. Pat. No. 5,288,514, entitled "Solid Phase and Combinatorial Synthesis of Benzodiazepine Compounds on a Solid Support," issued Feb. 22, 1994 and Bunin et al., "The Combinatorial Synthesis and Chemical and Biological Evaluation of a 1,4-Benzodiazepine Library," Proc. Natl. Acad. Sci. 91:4708-12 (1994). The structures of the individual library members can be decoded by analyzing the pin location in conjunction with the sequence of reaction steps used during the synthesis.
A second, related spatially-addressable strategy that has emerged involves solid-phase synthesis of polymers in individual reaction vessels, where the individual vessels are arranged into a single reaction unit. An illustrative example of such a reaction unit is a standard 96-well microtitre plate; the entire plate comprises the reaction unit and each well corresponds to a single reaction vessel. This approach is an extrapolation of traditional single-column solid-phase synthesis.
As is exemplified by the 96-well plate reaction unit, each reaction vessel is spatially defined by a two-dimensional matrix. Thus, the structures of individual library members can be decoded by analyzing the sequence of reactions to which each well was subjected.
Another spatially-addressable strategy employs "tea bags" to hold the synthesis resin. The reaction sequence to which each tea bag is subject is recorded, which determines the structure of the oligomer synthesized in each tea bag. See for example, Lam et al., "A New Type of Synthetic Peptide Library for Identifying Ligand-Binding Activity," Nature 354:82-84 (1991); Houghten et al., "Generation and Use of Synthetic Peptide Combinatorial Libraries for Basic Research and Drug Discovery," Nature 354:84-86 (1991); Houghten, "General Method for the Rapid Solid-Phase Synthesis of Large Numbers of Peptides: Specificity of Antigen-Antibody Interaction at the Level of Individual Amino Acids," Proc. Natl. Acad. Sci. 82:5131-5135 (1985); and Jung et al., Agnew. Chem. Int. Ed. Engl. 91:367-383 (1992), each of which is incorporated herein by reference.
In another recent development, scientists combined the techniques of photolithography, chemistry and biology to create large collections of oligomers and other compounds on the surface of a substrate (this method is called "VLSIPS.TM."). See, for example, U.S. Pat. No. 5,143,854; PCT Publication No. 90/15070; PCT Publication No. 92/10092 entitled "Very Large Scale Immobilized Polymer Synthesis," Jun. 25, 1992; Fodor et al., "Light-Directed Spatially Addressable Parallel Chemical Synthesis," Science 251:767-773 (1991); Pease et al., "Light-Directed Oligonucleotide Arrays for Rapid DNA Sequence Analysis," Proc. Natl. Acad. Sci. 91:5022-5026 (1994); and Jacobs & Fodor, "Combinatorial Chemistry: Applications of Light-Directed Chemical Synthesis," Trends. Biotechnology 12(1):19-26 (1994), each of which is incorporated herein by reference.
Others have developed recombinant methods for preparing collections of oligomers. See, for example, PCT Publication No. 91/17271; PCT Publication No. 91/19818; Scott, "Discovering Peptide Ligands Using Epitope Libraries," TIBS 17:241-245 (1992); Cwirla et al., "Peptides on Phage: A Vast Library of Peptides for Identifying Ligands," Proc. Natl. Acad. Sci. 87:6378-6382 (1990); Devlin et al., "Random Peptide Libraries: A Source of Specific Protein Binding Molecules," Science 249:404-406 (1990); and Scott & Smith, "Searching for Peptide Ligands with an Epitope Library," Science 249:386-390 (1990). Using these methods, one can identify each oligomer in the library by determining the coding sequences in the recombinant organism or phage. However, since the library members are generated in vivo, recombinant methods are limited to polymers whose synthesis is mediated in the cell. Thus, these methods typically have been restricted to constructing peptide libraries.
A third general strategy that has emerged involves the use of "split-bead" combinatorial synthesis strategies. See, for example, Furka et al., Int. J. Pept. Protein Res. 37:487-493 (1991), which is incorporated herein by reference. In this method synthesis supports are apportioned into aliquots, each aliquot exposed to a monomer, and the beads pooled. The beads are then mixed, reapportioned into aliquots, and exposed to a second monomer. The process is repeated until the desired library is generated.
Since the polymer libraries generated with the split-bead method are not spatially-addressable, the structures of the individual library members cannot be elucidated by analyzing the reaction histogram. Rather, structures must be determined by analyzing the polymers directly. Thus, one limitation of the split-bead approach is the requisite for an available means to analyze the polymer composition. While sequencing techniques are available for peptides and nucleic acids, sequencing reactions for polymers of other composition, such as for example carbohydrates, organics, peptide nucleic acids or mixed polymers may not be readily known.
Variations on the "split-bead" scheme have emerged that obviate the need to sequence the library member directly. These methods utilize chemicals to tag the growing polymers with a unique identification tag ("co-synthesis" strategies). See, for example, PCT Publication No. WO 94/08051 entitled "Complex Combinatorial Chemical Libraries Encoded with Tags," Apr. 14, 1994; Nestler et al., "A General Method for Molecular Tagging of Encoded Combinatorial Chemistry Libraries," J. Org. Chem. 59:4723-4724 (1994); PCT Publication No. WO 93/06121 entitled "Method of Synthesizing Diverse Collections of Oligomers," Apr. 1, 1993; Needels et al., Proc. Natl. Acad. Sci. 90:10700-10704 (1993); Kerr et al., "Encoded Combinatorial Peptide Libraries Containing Non-Natural Amino Acids," J. Amer. Chem. Soc. 115:2529-2531 (1993); and Brenner & Lerner, "Encoded Combinatorial Chemistry," Proc. Natl. Acad. Sci. 89:5381-5383 (1992), each of which is incorporated herein by reference.
Encoding library members with chemical tags occurs in such a fashion that unique identifiers of the chemical structures of the individual library members are constructed in parallel, or are co-synthesized, with the library members. Typically, in a linear three component synthesis containing building blocks A, B and C in the process of generating library member ABC, an encoding tag is introduced at each stage such that the tags T.sub.A, T.sub.B and T.sub.C would encode for individual inputs in the library. The synthesis would proceed as follows: (a) Chemical A is coupled onto a synthesis bead, immediately followed by coupling tag T.sub.A to the bead; (b) The bead is subject to deprotection conditions which remove the protecting group selectively from A, leaving T.sub.A protected. Chemical B is coupled to the bead, generating the sequence AB. The bead is then subject to deprotection which selectively removes the protecting group from T.sub.A, and T.sub.B is coupled to the bead, generating tag sequence T.sub.A T.sub.B ; (c) The third component C and concomitant tag T.sub.C is added to the bead in the manner described above, generating library sequence ABC and tag sequence T.sub.A T.sub.B T.sub.C.
For large libraries containing three chemical inputs, the chemical tagging sequence is the same. Thus, to generate a large library containing the complete set of three-input, one hundred unit length polymers, or 100.sup.3 =10.sup.6 library members, unique identifying tags are introduced such that there is a unique identifier tag for each different chemical structure. Theoretically, this method is applicable to libraries of any complexity as long as tagging sequences can be developed that have at least the same number of identification tags as there are numbers of unique chemical structures in the library.
While combinatorial synthesis strategies provide a powerful means for rapidly identifying target molecules, substantial problems remain. For example, since members of spatially addressable libraries must be synthesized in spatially segregated arrays, only relatively small libraries can be constructed. The position of each reaction vessel in a spatially-addressable library is defined by an XY coordinate pair such that the entire library is defined by a two-dimensional matrix. As the size of the library increases the dimensions of the two-dimensional matrix increases. In addition, as the number of different transformation events used to construct the library increases linearly, the library size increases exponentially. Thus, while generating the complete set of linear tetramers comprised of four different inputs requires only a 16.times.16 matrix (4.sup.4 =256 library members), generating the complete set of linear octamers composed of four different inputs requires a 256.times.256 matrix (4.sup.8 =65,536 library members), and generating the complete set of linear tetramers composed of twenty different inputs requires a 400.times.400 matrix (20.sup.4 =160,000 library members). Therefore, not only does the physical size of the library matrix quickly become unwieldy (constructing the complete set of linear tetramers composed of twenty different inputs using spatially-addressable techniques requires 1667 microtitre plates), delivering reagents to each reaction vessel in the matrix requires either tedious, time-consuming manual manipulations, or complex, expensive automated equipment.
While the VLSIPS.TM. method attempts to overcome this limitation through miniaturization, VLSIPS.TM. requires specialized photoblocking chemistry, expensive, specialized synthesis equipment and expensive, specialized assay equipment. Thus, the VLSIPS.TM. method is not readily and economically adaptable to emerging solid phase chemistries and assay methodologies.
Split bead methods also suffer severe limitations. Although large libraries can theoretically be constructed using split-bead methods, the identity of library members displaying a desirable property must be determined by analytical chemistry. Accordingly, split-bead methods can only be employed to synthesize compounds that can be readily elucidated by microscale sequencing, such as polypeptides and polynucleotides.
Co-synthesis strategies have attempted to solve this structure elucidation problem. However, these methods also suffer limitations. For example, the tagging structures may be incompatible with synthetic organic chemistry reagents and conditions. Additional limitations follow from the necessity for compatible protecting groups which allow the alternating co-synthesis of tag and library member, and assay confusion that may arise from the tags selectively binding to the assay receptor.
Finally, since methods such as the preceding typically require the addition of like moieties, there is substantial interest in discovering methods for producing labeled libraries of compounds which are not limited to sequential addition of like moieties, and which are amenable to any chemistries now known or that will be later developed to generate chemical libraries. Such methods would find application, for example, in the modification of steroids, sugars, co-enzymes, enzyme inhibitors, ligands and the like, which frequently involve a multi-stage synthesis in which one would wish to vary the reagents and/or conditions to provide a variety of compounds.
In such methods the reagents may be organic or inorganic reagents, where functionalities or side groups may be introduced, removed or modified, rings opened or closed, stereochemistry changed, and the like.
From the above, one can recognize that there is substantial interest in developing improved methods and apparatus for the synthesis of complex labeled combinatorial chemical libraries which readily permit the construction of libraries of virtually any composition and which readily permit accurate structural determination of individual compounds within the library that are identified as being of interest. Many of the disadvantages of the previously-described methods as well as many of the needs not met by them are addressed by the present invention, which as described more fully hereinafter, provides myriad advantages over these previously-described methods.